An online labeling quality detection system and method

Abstract

This invention relates to the field of visual inspection technology and discloses an online labeling quality inspection system and method, including a spatial coding light projection module, an image acquisition module, a light intensity pre-regulation unit, and a signal processing module. The light intensity pre-regulation unit determines the conjugate modulation site of the specular reflection region based on the imaging and projection optical center and the transmission trajectory of the tested container. By superimposing an asymmetric spatial energy flow attenuation field at this site, the incident brightness is pre-controlled, so that the charge accumulation of the reflected image is maintained within the full-well charge capacity limit. The signal processing module analyzes the phase information and reconstructs the three-dimensional morphological features of the labeling region to establish evaluation indicators. This invention suppresses high-light interference before interaction, eliminates cross-frame adjustment lag, realizes feature extraction and three-dimensional reconstruction of labeling defects against a strongly reflective curved surface background, and improves the detection accuracy under high-speed flow conditions.

Description

Technical Field

[0001] This invention belongs to the field of visual inspection technology, and in particular relates to an online labeling quality inspection system and method. Background Technology

[0002] Currently, label quality inspection in automated production lines mainly adopts visual recognition. Existing technology uses a light source to illuminate the surface of the object under test, and the sensor acquires the reflected light signal from the label surface. Defects are determined based on edge contours or grayscale features. Visual inspection relies on the interaction between the probe light field and the surface of the object under test. The light beam generates a modulation effect at the target topological interface, and the sensor collects the distorted light field signal carrying morphological features to restore the surface micro-state. The pharmaceutical and daily chemical industries have increased requirements for the light transmittance and gloss of packaging containers. The superposition of transparent labels and high-curvature reflective containers causes specular reflection of the probe light field at the multi-layer physical interface. Under high-speed flow conditions, the reflection generatrix generated by the curved surface of the container causes pixel overflow, resulting in highlight areas in the image. Although shortening the exposure time or reducing the light intensity avoids saturation overflow, it leads to a decrease in the signal-to-noise ratio of weak contrast defects.

[0003] Industry experts have attempted to broaden the detection dynamic range by employing multiple exposure methods or constructing brightness closed-loop feedback logic. However, in production lines with short sampling cycles, multiple exposures cause motion blur, damaging high-frequency phase information at the edges of the morphology. Brightness adjustment based on image feedback suffers from cross-frame response delays. When compensation parameters determined for the previous test object are applied to subsequent test objects, random shifts in the test object's pose cause mismatches in the compensation area, making it impossible to suppress specular interference at its physical source. In addition to the hardware layout limitations caused by the high curvature of the container surface, related software control methods face constraints from underlying mechanisms. For example, Chinese invention patent with publication number CN121437511A... The patent application discloses a system and method for detecting appearance defects based on CCD image recognition. It restores surface features by identifying specular reflection interference areas and combining texture restoration technology. This scheme relies on grayscale gradient information driven by the backend Poisson fusion or gradient filling operation under the premise that the imaging chip has not experienced saturation overflow. When processing packaging containers with high curvature and specular surface characteristics, the instantaneous reflection energy far exceeds the full-well capacity of the photosensitive chip, resulting in physical feature truncation in the local image. The algorithm cannot achieve true reconstruction of micro-defects due to the loss of initial phase constraints. The image restoration algorithm involves a large number of iterative calculations, which can easily lead to computational stack overflow within the millisecond-level production line sampling cycle.

[0004] Therefore, how to acquire and reconstruct the topological features of transparent labels under high-speed circulation and specular reflection conditions has become the technical problem to be solved by this invention. Summary of the Invention

[0005] This invention provides an online labeling quality inspection system for detecting the labeling quality of a cylindrical container under test during transmission, comprising: A spatially encoded optical projection module is used to project a dual-frequency modulated light field with sinusoidal grayscale distribution characteristics onto the surface of a cylindrical container under test. The image acquisition module is used to acquire the modulated reflection image of the surface of the cylindrical container under test; The light intensity pre-regulation unit has its input end connected to the signal of the image acquisition module and its output end connected to the spatial coding light projection module. Based on the imaging optical center of the image acquisition module, the projection optical center of the spatial coding light projection module, and the transmission trajectory of the cylindrical container under test, the light intensity pre-regulation unit determines the conjugate modulation site of the specular reflection area on the spatial coding light projection module. When generating the brightness driving matrix of the dual-frequency modulation light field, the light intensity pre-regulation unit superimposes a static asymmetric spatial energy flow attenuation field in the local image domain where the conjugate modulation site is located. Before the photons interact with the surface of the cylindrical container under test, the incident brightness flux of the specular reflection area is reduced in advance, so that the charge accumulation of the highly reflective area in the modulated reflection image is maintained within the full-well charge capacity limit set by the photosensitive chip of the image acquisition module. The signal processing module is used to analyze the phase information of the modulated reflection image, extract the high-frequency distortion component that characterizes the labeling defects, and use the low-frequency wrapping phase distribution in the dual-frequency modulated light field as the order reference to complete the spatial phase unfolding of the high-frequency wrapping phase distribution, reconstruct the three-dimensional morphological features of the labeling area, and establish the labeling quality evaluation index by comparing the three-dimensional morphological features with the preset geometric model.

[0006] Preferably, the system further includes a hardware phase-locked loop; the hardware phase-locked loop is connected to the displacement pulse output terminal of the transmission device and is used to lock the trigger pulse of the image acquisition module and the displacement pulse of the cylindrical container under test; the signal processing module calculates the spatial coordinate offset of the cylindrical container under test when the phase difference stability parameter deviates from the preset range based on the phase difference stability parameter between adjacent frame modulated reflection images, and performs coordinate compensation for the conjugate modulation site determined by the light intensity pre-regulation unit.

[0007] Preferably, the dual-frequency modulated optical field includes a high-frequency sinusoidal component and a low-frequency sinusoidal component, with a spatial frequency ratio of 10 to 30. The signal processing module uses the low-frequency phase field generated by the low-frequency sinusoidal component to establish the fringe order generated by the high-frequency sinusoidal component on the surface of the tested cylindrical container, and performs unwrapping operation on the phase field generated by the high-frequency sinusoidal component to eliminate the phase unwrapping deviation caused by the periodic jump of the fringe at the edge of the tested cylindrical container with high curvature.

[0008] Preferably, the spatially encoded light projection module includes a spatial light modulator composed of digital micromirror devices; the light intensity pre-regulation unit discretizes the two-dimensional coordinates of the conjugate modulation site on the spatial light modulator to generate a grayscale correction matrix corresponding to the static asymmetric spatial energy flow attenuation field, and superimposes the grayscale correction matrix onto the bottom driving sequence of the dual-frequency modulated light field, so that the projection energy of the mirror reflection area is non-uniformly distributed according to the curvature gradient of the surface of the tested cylindrical container.

[0009] Preferably, when reconstructing the three-dimensional morphological features, the signal processing module extracts the energy feature distribution of the modulated reflection image in the spatial frequency domain; when there are microbubble defects on the label on the surface of the tested cylindrical container, the signal processing module identifies the local frequency offset signal in the modulated reflection image that deviates from the preset frequency threshold, thereby achieving physical isolation between the structural component corresponding to the microbubble defect and the background noise of the specular reflection area on the surface of the tested cylindrical container in the spatial frequency domain.

[0010] Preferably, the image acquisition module includes an industrial camera and a fixed-focus optical lens. The acquisition frequency of the industrial camera is not less than 100Hz. The main optical axis of the industrial camera and the projection axis of the spatial coding light projection module are arranged at an angle of 30 to 60 degrees in space, so that the imaging target surface of the image acquisition module and the modulation plane of the spatial coding light projection module form an optical conjugate mapping relationship, so as to establish a spatial sampling grid for the surface of the cylindrical container under test.

[0011] Preferably, the signal processing module calculates the gradient deviation of the three-dimensional topographic features in the tangential direction of the surface; when the gradient deviation continues to exceed the preset slope threshold at the label edge, the signal processing module generates a judgment instruction characterizing the label edge lifting defect, and uses the local normal displacement of the reconstructed label area on the surface of the tested cylindrical container as a quantitative indicator of the defect degree to identify the label peeling state of the tested cylindrical container.

[0012] Preferably, the image acquisition module is equipped with a linear polarization filter on the light-incident side; the signal processing module performs temporal smoothing filtering on 3 to 5 consecutive frames of modulated reflection images to reduce the signal-to-noise ratio loss caused by the mechanical displacement vibration of the tested cylindrical container during high-speed transmission, and suppresses the random phase drift caused by ambient stray light, so as to improve the measurement repeatability of three-dimensional morphological features.

[0013] Preferably, the system further includes a defective product separation unit; the defective product separation unit is connected to the output of the signal processing module and is used to receive the judgment instruction generated by the signal processing module. When the three-dimensional morphological features do not meet the preset quality standards, a physical rejection action is generated to remove the target cylindrical container being tested from the main transmission path.

[0014] An online labeling quality detection method includes the following steps: Based on the imaging optical center of the image acquisition module, the projection optical center of the spatial coding optical projection module, and the transmission trajectory of the cylindrical container under test, the conjugate modulation site of the specular reflection area on the surface of the cylindrical container under test on the spatial coding optical projection module is determined. When generating the brightness driving matrix of the dual-frequency modulated light field, a static asymmetric spatial energy flux attenuation field is superimposed on the local image domain where the conjugate modulation site is located. Before the photon interacts with the surface of the cylindrical container under test, the incident brightness flux of the specular reflection area is reduced in advance, so that the charge accumulation in the highly reflective area of ​​the modulated reflection image is maintained within the full-well charge capacity limit of the photosensitive chip. The trigger pulse of the image acquisition module and the displacement pulse of the cylindrical container under test are locked by a hardware phase-locked loop to obtain the modulated reflection image of the surface of the cylindrical container under test. The phase information of the modulated reflection image is analyzed, the high-frequency distortion component characterizing the labeling defect is extracted, and the low-frequency wrapping phase distribution in the dual-frequency modulated light field is used as the order reference to complete the spatial phase unfolding of the high-frequency wrapping phase distribution and reconstruct the three-dimensional morphological features of the labeling area. Calculate the gradient deviation of the three-dimensional topographic features in the tangential direction of the surface. When the gradient deviation continues to exceed the preset slope threshold at the label edge, generate a judgment command characterizing the label edge lifting defect. Upon receiving the judgment instruction, if the three-dimensional morphological features do not meet the preset quality standards, the non-conforming product separation unit generates a physical rejection action to remove the target cylindrical container from the main transmission path.

[0015] Compared with existing technologies, the online labeling quality detection system of the present invention has the following advantages: 1. In online labeling quality inspection, by superimposing a dual-frequency sinusoidal modulated light field with a carrier frequency in the spatial domain, the local geometric abrupt changes caused by tiny defects on the surface of the test object are converted into spatial frequency broadening and phase shift in the reflected light field. This mechanism of extracting features in the spatial frequency dimension enables the weak deformation signal generated by microbubbles to be physically isolated from the low-frequency DC background component represented by the specular reflection spot in the frequency domain, thereby improving the detection signal-to-noise ratio of transparent labels against a highly reflective curved surface background.

[0016] 2. The pre-modulation logic module, based on the optical axis center of the image acquisition unit, the projection optical center of the spatial coding optical field projection unit, and the three-dimensional spatial pose relationship of the cylindrical container under test, superimposes a static spatial energy attenuation envelope onto the dual-frequency sinusoidal modulated light field before the photons interact with the object under test. This energy envelope, based on the inverse mapping of the reflectivity of the cylindrical curved surface mirror, pre-compresses the optical energy flux of the high-reflectivity region, forcibly limiting the charge accumulation of the sensor pixel to below the maximum full-well charge capacity limit of the detector pixel, avoiding cross-frame time lag caused by image feedback adjustment under high-speed pipeline conditions, and maintaining the fringe topological continuity required for global absolute phase reconstruction.

[0017] 3. Using the low-frequency wrapped phase distribution as the order reference, spatial phase unfolding is performed on the high-frequency wrapped phase distribution to solve the phase unpacking error caused by the stripe periodic jump at the edge of the high curvature bottle body; this mechanism ensures the reconstruction accuracy of the system for micron-level edge warping defects through the phase state coordination of multi-frequency components, and reduces the phase order jump and unpacking logic blind zone caused by geometric abrupt changes. Attached Figure Description

[0018] Figure 1 This is the overall logical architecture and processing flowchart of the labeling quality online detection system of the present invention; Figure 2 This is a schematic diagram illustrating the hardware and software collaborative operation principle of the online labeling quality detection system of this invention. Detailed Implementation

[0019] The technical solutions of the embodiments of this application will be clearly described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of this application are within the scope of protection of this application.

[0020] It should be noted that all directional and positional terms used in this invention, such as: up, down, left, right, front, back, vertical, horizontal, inner, outer, top, bottom, transverse, longitudinal, center, etc., are only used to explain the relative positional relationship and connection between components in a specific state (as shown in the accompanying drawings). They are only for the convenience of describing this invention and do not require that this invention be constructed and operated in a specific orientation. Therefore, they should not be construed as limiting this invention. In addition, the descriptions of "first," "second," etc., in this invention are for descriptive purposes only and should not be construed as indicating or implying their relative importance or implicitly specifying the number of technical features indicated.

[0021] In the description of this invention, unless otherwise explicitly specified and limited, the terms installation, connection, and linking should be interpreted broadly. For example, they can refer to fixed connections, detachable connections, or integral connections; they can refer to mechanical connections; they can refer to direct connections or indirect connections through an intermediate medium; they can refer to the internal connection of two components. For those skilled in the art, the specific meaning of the above terms in this invention can be understood according to the specific circumstances.

[0022] In the description of this specification, references to the terms "an embodiment," "some embodiments," "illustrative embodiments," "examples," "specific examples," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the present invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example, and the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0023] An online labeling quality inspection system is used to inspect the labeling quality of a cylindrical container under test during transmission, comprising: A spatially encoded optical projection module is used to project a dual-frequency modulated light field with sinusoidal grayscale distribution characteristics onto the surface of a cylindrical container under test. The image acquisition module is used to acquire the modulated reflection image of the surface of the cylindrical container under test; The light intensity pre-regulation unit has its input end connected to the signal of the image acquisition module and its output end connected to the spatial coding light projection module. Based on the imaging optical center of the image acquisition module, the projection optical center of the spatial coding light projection module, and the transmission trajectory of the cylindrical container under test, the light intensity pre-regulation unit determines the conjugate modulation site of the specular reflection area on the spatial coding light projection module. When generating the brightness driving matrix of the dual-frequency modulation light field, the light intensity pre-regulation unit superimposes a static asymmetric spatial energy flow attenuation field in the local image domain where the conjugate modulation site is located. Before the photons interact with the surface of the cylindrical container under test, the incident brightness flux of the specular reflection area is reduced in advance, so that the charge accumulation of the highly reflective area in the modulated reflection image is maintained within the full-well charge capacity limit set by the photosensitive chip of the image acquisition module. The signal processing module is used to analyze the phase information of the modulated reflection image, extract the high-frequency distortion component that characterizes the labeling defects, and use the low-frequency wrapping phase distribution in the dual-frequency modulated light field as the order reference to complete the spatial phase unfolding of the high-frequency wrapping phase distribution, reconstruct the three-dimensional morphological features of the labeling area, and establish the labeling quality evaluation index by comparing the three-dimensional morphological features with the preset geometric model.

[0024] Preferably, the system further includes a hardware phase-locked loop; the hardware phase-locked loop is connected to the displacement pulse output terminal of the transmission device and is used to lock the trigger pulse of the image acquisition module and the displacement pulse of the cylindrical container under test; the signal processing module calculates the spatial coordinate offset of the cylindrical container under test when the phase difference stability parameter deviates from the preset range based on the phase difference stability parameter between adjacent frame modulated reflection images, and performs coordinate compensation for the conjugate modulation site determined by the light intensity pre-regulation unit.

[0025] Preferably, the dual-frequency modulated optical field includes a high-frequency sinusoidal component and a low-frequency sinusoidal component, with a spatial frequency ratio of 10 to 30. The signal processing module uses the low-frequency phase field generated by the low-frequency sinusoidal component to establish the fringe order generated by the high-frequency sinusoidal component on the surface of the tested cylindrical container, and performs unwrapping operation on the phase field generated by the high-frequency sinusoidal component to eliminate the phase unwrapping deviation caused by the periodic jump of the fringe at the edge of the tested cylindrical container with high curvature.

[0026] Preferably, the spatially encoded light projection module includes a spatial light modulator composed of digital micromirror devices; the light intensity pre-regulation unit discretizes the two-dimensional coordinates of the conjugate modulation site on the spatial light modulator to generate a grayscale correction matrix corresponding to the static asymmetric spatial energy flow attenuation field, and superimposes the grayscale correction matrix onto the bottom driving sequence of the dual-frequency modulated light field, so that the projection energy of the mirror reflection area is non-uniformly distributed according to the curvature gradient of the surface of the tested cylindrical container.

[0027] Preferably, when reconstructing the three-dimensional morphological features, the signal processing module extracts the energy feature distribution of the modulated reflection image in the spatial frequency domain; when there are microbubble defects on the label on the surface of the tested cylindrical container, the signal processing module identifies the local frequency offset signal in the modulated reflection image that deviates from the preset frequency threshold, thereby achieving physical isolation between the structural component corresponding to the microbubble defect and the background noise of the specular reflection area on the surface of the tested cylindrical container in the spatial frequency domain.

[0028] Preferably, the image acquisition module includes an industrial camera and a fixed-focus optical lens. The acquisition frequency of the industrial camera is not less than 100Hz. The main optical axis of the industrial camera and the projection axis of the spatial coding light projection module are arranged at an angle of 30 to 60 degrees in space, so that the imaging target surface of the image acquisition module and the modulation plane of the spatial coding light projection module form an optical conjugate mapping relationship, so as to establish a spatial sampling grid for the surface of the cylindrical container under test.

[0029] Preferably, the signal processing module calculates the gradient deviation of the three-dimensional topographic features in the tangential direction of the surface; when the gradient deviation continues to exceed the preset slope threshold at the label edge, the signal processing module generates a judgment instruction characterizing the label edge lifting defect, and uses the local normal displacement of the reconstructed label area on the surface of the tested cylindrical container as a quantitative indicator of the defect degree to identify the label peeling state of the tested cylindrical container.

[0030] Preferably, the image acquisition module is equipped with a linear polarization filter on the light-incident side; the signal processing module performs temporal smoothing filtering on 3 to 5 consecutive frames of modulated reflection images to reduce the signal-to-noise ratio loss caused by the mechanical displacement vibration of the tested cylindrical container during high-speed transmission, and suppresses the random phase drift caused by ambient stray light, so as to improve the measurement repeatability of three-dimensional morphological features.

[0031] Preferably, the system further includes a defective product separation unit; the defective product separation unit is connected to the output of the signal processing module and is used to receive the judgment instruction generated by the signal processing module. When the three-dimensional morphological features do not meet the preset quality standards, a physical rejection action is generated to remove the target cylindrical container being tested from the main transmission path.

[0032] An online labeling quality detection method includes the following steps: Based on the imaging optical center of the image acquisition module, the projection optical center of the spatial coding optical projection module, and the transmission trajectory of the cylindrical container under test, the conjugate modulation site of the specular reflection area on the surface of the cylindrical container under test on the spatial coding optical projection module is determined. When generating the brightness driving matrix of the dual-frequency modulated light field, a static asymmetric spatial energy flux attenuation field is superimposed on the local image domain where the conjugate modulation site is located. Before the photon interacts with the surface of the cylindrical container under test, the incident brightness flux of the specular reflection area is reduced in advance, so that the charge accumulation in the highly reflective area of ​​the modulated reflection image is maintained within the full-well charge capacity limit of the photosensitive chip. The trigger pulse of the image acquisition module and the displacement pulse of the cylindrical container under test are locked by a hardware phase-locked loop to obtain the modulated reflection image of the surface of the cylindrical container under test. The phase information of the modulated reflection image is analyzed, the high-frequency distortion component characterizing the labeling defect is extracted, and the low-frequency wrapping phase distribution in the dual-frequency modulated light field is used as the order reference to complete the spatial phase unfolding of the high-frequency wrapping phase distribution and reconstruct the three-dimensional morphological features of the labeling area. Calculate the gradient deviation of the three-dimensional topographic features in the tangential direction of the surface. When the gradient deviation continues to exceed the preset slope threshold at the label edge, generate a judgment command characterizing the label edge lifting defect. Upon receiving the judgment instruction, if the three-dimensional morphological features do not meet the preset quality standards, the non-conforming product separation unit generates a physical rejection action to remove the target cylindrical container from the main transmission path.

[0033] Example 1: In an automated labeling production line for transparent containers with a turnover rate of 100 bottles / min or higher, the high curvature surface of the cylindrical container being tested, combined with the high transmittance label, causes the probe beam to be mirrored at the interface. Traditional visual inspection systems suppress the overflow of local pixels by using a closed-loop dynamic dimming of the light source based on hysteresis image calculation. However, relying on a post-compensation mechanism in the control feedback domain, cross-frame physical misalignment is caused within the exposure window of 1μs to 999μs, resulting in the truncation of high-frequency optical features characterizing bubble defects of 1μm to 999μm in size. The labeling quality online inspection system described in this application places the high-gloss suppression in the physical spatial configuration domain before the photon interacts with the surface being tested, and implements spatial energy pre-shaping based on static geometric topology to avoid physical timing conflicts between high-speed turnover conditions and dynamic adjustment of the light field.

[0034] The light intensity pre-regulation unit, based on the imaging optical center of the image acquisition module, the projection optical center of the spatial coding light projection module, and the transmission trajectory coordinates of the tested cylindrical container, resolves the conjugate modulation site of the specular reflection area on the surface of the tested cylindrical container on the spatial coding light projection module. When generating the brightness driving matrix of the dual-frequency modulation light field with sinusoidal gray-scale distribution characteristics, a static asymmetric spatial energy flux attenuation field is superimposed on the local image domain where the conjugate modulation site is located. This causes the optical energy flux projected onto the highly reflective area to undergo non-uniform reduction in the physical spatial domain according to the surface reflectivity distribution. This asymmetric spatial energy flux attenuation field cuts off the overexposure blind zone generation path, keeping the charge accumulation in the highly reflective area of ​​the modulated reflection image acquired by the image acquisition module within the full-well charge capacity limit of the photosensitive chip. This maintains the continuity and integrity of the spatial fringe topology and provides an uninterrupted basic phase source for the signal processing module, enabling the signal processing module to extract the high-frequency distortion component characterizing the labeling defect. At the same time, the low-frequency wrapping phase distribution in the dual-frequency modulation light field is used as a level reference to complete the spatial phase unfolding of the high-frequency wrapping phase distribution.

[0035] With the trigger pulse of the image acquisition module and the displacement pulse of the tested cylindrical container physically locked at equal spatial intervals via a hardware phase-locked loop, the signal processing module separates background noise and structured light features in the spatial frequency domain using bandpass filtering. It calculates the gradient deviation of the three-dimensional morphological features obtained after spatial phase unfolding in the tangential direction of the curved surface and uses the local normal displacement of the reconstructed labeling area on the surface of the tested cylindrical container as a judgment parameter. Within an exposure time window of 1μs to 999μs, it achieves physical isolation between the structural components corresponding to bubbles and edge lifting defects and strong reflective background noise in the spatial frequency domain. Under a highly reflective curved surface background, it reconstructs the three-dimensional morphological features of the labeling area and establishes a labeling quality evaluation index that includes quantitative data on the degree of defects.

[0036] Example 2: This example sets up an automated labeling production line for cylindrical containers under test with a conveying cycle of 300 bottles / min to verify the detection capability of labeling defects in the cylindrical containers under test. The physical experimental platform on which the verification data is collected includes a global shutter industrial camera with a sampling frequency of 200Hz and a spatially encoded light projection module in the form of a digital micromirror array. A spatial angle is set between the main optical axis and the projection axis of the spatially encoded light projection module to balance the optical resolution at the edge of the imaging field of view and the direct light flux in the specular reflection area. When the radius of curvature of the cylindrical container under test is less than 20mm, according to the geometric reflection law, reducing the angle increases the direct energy impact in the optical axis normal direction, while expanding the angle causes perspective distortion in the sampling grid in the edge area of ​​the field of view. Therefore, a reference verification state of 45° is set for the main axis angle to balance the field of view coverage and high light suppression requirements. A 50Hz illuminance fluctuation is injected into the ambient lighting source, and Gaussian white noise with a signal-to-noise ratio of 20dB is superimposed in the transmission loop of the image acquisition module to simulate photoelectric disturbances in the industrial environment. A control group with missing features was established. This control group used conventional dual-frequency modulated light field to extract three-dimensional morphological features. The extracted raw image data showed that under the interference of ambient illumination fluctuations and Gaussian white noise, the charge accumulation of the specular reflection band on the surface of the tested cylindrical container reached the full-well limit of 65535 ADU of the 16-bit photosensitive chip. The local charge overflow of the photosensitive chip caused the spatial frequency domain topology truncation of the modulated reflection image, with a truncation width of 450 μm. The sample group of the present invention was constructed, and the light intensity pre-regulation unit was activated. Based on the imaging optical center of the image acquisition module, the projection optical center of the spatial coding light projection module, and the transmission trajectory coordinates of the tested cylindrical container, the light intensity pre-regulation unit resolved the conjugate modulation site of the specular reflection region on the spatial coding light projection module, and superimposed a static asymmetric spatial energy flow attenuation field on the local image domain where it was located. The acquired modulated reflection image showed that the charge accumulation of the conjugate modulation site was reduced to 42105 ADU. The asymmetric spatial energy flow attenuation field cut off the photon accumulation path of the overexposed area and maintained the spatial fringe topology continuity and integrity of the dual-frequency modulated light field.

[0037] Sample groups and an out-of-range control group with different principal optical axis intersection parameters were constructed to verify the reconstruction parameter gradient of the spatial sampling grid. The principal optical axis intersection angles of sample group 1, sample group 2, and sample group 3 were set to 30°, 45°, and 60°, respectively, and the principal optical axis intersection angles of out-of-range control group 1 and out-of-range control group 2 were set to 20° and 75°, respectively. The signal processing module separated background noise by bandpass filtering in the spatial frequency domain, expanded the spatial phase of the high-frequency wrapped phase distribution using the low-frequency wrapped phase distribution as the order benchmark, and calculated the gradient deviation of the three-dimensional topographic features in the tangential direction of the surface. The extracted effective three-dimensional point cloud density data showed that the point cloud density of sample group 1 was 152 points / mm², and the point cloud density of sample group 2 was... The point cloud density of sample group three of this invention reached 218 points / mm², while that of sample group three was 185 points / mm². When the angle of the principal optical axis reached 20°, which is the threshold of the asymmetric spatial energy flow attenuation field, the direct light penetrated the asymmetric spatial energy flow attenuation field, and the point cloud density dropped to 45 points / mm². When the angle reached 75°, which is the threshold of the control group two, the edge geometric distortion caused the point cloud density to drop to 62 points / mm². The above point cloud density data showed a nonlinear parabolic trend with the change of the angle. The 30° to 60° range was established as the engineering operation window for separating high-frequency defects and maintaining the integrity of the field of view. The recognition accuracy of each group for microbubble defects with a size of 15μm was statistically analyzed. The recognition accuracy of the partially missing control group was 42.5%, while the recognition accuracy of sample group two of this invention increased to 99%.2%, the asymmetric spatial energy flow attenuation field pre-shaping of the physical spatial domain of strong reflection energy coordinates with the signal processing module's high-frequency distortion component extraction mechanism in the spatial frequency domain; this static energy pre-carving provides the signal processing module with an untruncation basic phase source, eliminates cross-frame physical misalignment caused by dynamic image feedback logic, reconstructs the three-dimensional morphological features of the labeling area against a strongly reflective curved surface background, and establishes an online evaluation index for labeling quality that includes local normal displacement. According to the spatial sampling theorem, to achieve frequency domain decoupling between the curved background and minor defects, the spatial frequency ratio of the high-frequency sinusoidal component to the low-frequency sinusoidal component of the dual-frequency modulated light field is set to 10 to 30. The upper and lower limits of this ratio are derived from the extreme value law of optical geometric sampling. When the frequency ratio is less than 10, the actual projected pixel spacing of the high-frequency stripes will be excessively amplified and cover the typical physical size limit of the microbubble, causing the weak phase distortion information to be completely hidden in the low-frequency carrier and lose decoupling. Resolution; however, when this frequency ratio exceeds the threshold upper limit of 30, the maximum curvature edge projection of the tested cylindrical surface will cause the dense high-frequency stripe period to be distorted and compressed below the physical limit of a single photosensitive pixel, thereby causing spatial aliasing of the optical signal and leading to global phase jump failure. The spatial period of the low-frequency sinusoidal component is set to be greater than the reference outer diameter of the tested cylindrical container to ensure that the single-cycle low-frequency light field completely covers the projection area of ​​the container surface to construct continuous substrate morphology information; the spatial period of the high-frequency sinusoidal component is constrained within a range of 2 to 5 times the feature size of the microbubble defect to meet the high-frequency sampling requirements of the defect contour. During the spatial phase unfolding process, the signal processing module performs a two-dimensional fast Fourier transform on the modulated reflection image to obtain the spatial energy spectrum. A band-stop filter is used to filter out the substrate DC component that matches the spatial frequency of the low-frequency sinusoidal component, stripping the curvature fluctuations of the tested cylindrical container from the total phase, and extracting the three-dimensional point cloud of the label's local deformation after removing substrate geometric disturbances.

[0038] Example 3: This example combines Figures 1 to 2 A description of an online labeling quality detection system and method, such as... Figure 1As shown, in a scenario where the tested cylindrical container is in a transmission state and its surface includes labeling and specular reflection areas, a dual-frequency modulated light field with sinusoidal grayscale distribution characteristics is projected onto the surface of the tested cylindrical container by a spatial coding light projection module. Simultaneously, an image acquisition module acquires the modulated reflection image of the tested cylindrical container surface. Through a signal connection that provides the imaging optical center and transmission trajectory input, a light intensity pre-regulation unit determines the conjugate modulation site and superimposes a static asymmetric spatial energy flow attenuation field to maintain the charge accumulation in the high-reflectivity area within the full-well charge capacity limit. The brightness driving matrix of the superimposed static asymmetric spatial energy flow attenuation field is fed back to the spatial coding light projection module. At the same time, the signal processing module receives the transmitted modulated reflection image and performs phase information analysis and high-frequency distortion component extraction. It performs spatial phase unfolding of the high-frequency wrapped phase distribution to reconstruct the three-dimensional morphological features of the labeling area and outputs an information stream containing morphological features and judgment data. Finally, the labeling quality evaluation index is established based on the comparison between the three-dimensional morphological features and the preset geometric model.

[0039] like Figure 2 As shown, in the system's interaction and processing flow, the transmission device outputs pulse signals to lock the trigger pulse and displacement pulse. The spatial homography matrix H is calculated by establishing the basic spatial mapping and reference threshold parameters, and an asymmetric spatial energy flow attenuation field is superimposed to suppress high-light interference. At the same time, the conjugate modulation site is determined so that after the system projects the dual-frequency modulated light field onto the cylindrical container under test and obtains the modulated reflection image, it can successfully complete the spatial phase unfolding and reconstruct the three-dimensional morphological features of the labeling area. Then, the high-frequency distortion component characterizing the labeling defect is extracted. When the gradient data exceeds the preset slope threshold, the system establishes the labeling quality evaluation index and generates a judgment command, which finally drives the non-conforming product separation unit to complete the processing of the target container.

[0040] Example 4: Before the automated labeling production line of the tested cylindrical container is put into operation, the labeling quality online detection system establishes the basic spatial mapping and reference threshold parameters. The light intensity pre-regulation unit drives the spatial coding light projection module to project an initial uniform dual-frequency modulated light field onto the standard cylindrical container without labeling defects. The image acquisition module acquires an initial reference image containing specular reflection features and extracts the coordinates of saturated pixel groups whose gray values ​​reach the full-well limit in the initial reference image. The light intensity pre-regulation unit inversely maps the coordinates of the saturated pixel groups to the physical coordinate system of the digital micromirror array of the spatial coding light projection module according to the spatial homography matrix, thereby establishing the conjugate of the specular reflection region. Modulation site; For the conjugate modulation site, the light intensity pre-regulation unit extracts the gray-level attenuation gradient from the center to the edge of the saturated pixel group in the initial reference image, and generates a brightness driving matrix that is inversely proportional to the gray-level attenuation gradient. The spatial coding light projection module reduces the initial projection energy in the central region of the conjugate modulation site according to the brightness driving matrix, while smoothly increasing the projection energy flux to the periphery, thereby forming a static asymmetric spatial energy flow attenuation field. The asymmetric spatial energy flow attenuation field causes the peak charge accumulation of the specular reflection band in the initial reference image to converge to 80% of the full-well capacity of the photosensitive chip. The light intensity pre-regulation unit solidifies the brightness driving matrix as the pre-regulation basis.

[0041] The signal processing module extracts defect judgment parameters, and the image acquisition module continuously acquires modulated reflection images of 50 standard cylindrical containers. It uses a two-dimensional fast Fourier transform to extract the fundamental frequency distribution energy spectrum of the modulated reflection images in the spatial frequency domain. The signal processing module calculates the spatial statistical variance of the 50 fundamental frequency distribution energy spectra and defines the sum of the energy spectrum mean and three times the spatial statistical variance as a preset frequency threshold in the spatial frequency domain. For the label edge lifting defect, the signal processing module extracts the reference normal gradient of the surface tangent direction in the label edge region of the standard cylindrical containers. The signal processing module calculates the upper limit of the feature envelope of the 50 reference normal gradients and simultaneously sets the upper limit of the envelope... The absolute scalar value is fixed as a preset slope threshold. In the specific isolation operation, the signal processing module calculates the corresponding high-frequency response range in the frequency domain based on the standard equivalent outer diameter range of the microbubble defect (15μm to 50μm) through the spatial sampling frequency conversion formula. In this way, a two-dimensional band-stop digital filter with a specific cutoff frequency parameter is constructed and applied to the acquired spatial energy spectrum. The low-frequency response belonging to the continuous surface of the substrate and the high-frequency shot noise deviating from the extreme value are accurately filtered and cut off, thereby completing the independent extraction of the microbubble characteristic frequency band. The labeling quality online detection system relies on the fixed threshold parameter to identify the local frequency offset signal in the modulated reflection image that deviates from the preset frequency threshold.

[0042] Example 5: In the debugging scenario of deploying the online labeling quality inspection system on a pharmaceutical vial production line, by attaching a calibration piece with a standard checkerboard texture to the surface of the cylindrical container being tested, the physical alignment of the image space and the projection space is completed. The image acquisition module obtains the pixel coordinate sequence containing the calibration feature points, and the light intensity pre-regulation unit extracts the feature point set in the pixel coordinate system. and and the corresponding driving matrix coordinate set in the spatial coding optical projection module and The least squares method is then applied to solve the system of linear equations, thereby determining the spatial homography matrix that characterizes the projective transformation relationship between the imaging plane and the modulation plane. The light intensity pre-regulation unit will control the spatial homography matrix. It is stored in the non-volatile memory area of ​​the system memory and used as the physical mapping basis for resolving the conjugate modulation sites in the subsequent detection process.

[0043] When the surface reflectivity of the tested cylindrical container fluctuates with batches of container materials, the system performs incident energy flow gain calibration based on the measured value of reflection intensity. The spatial coding light projection module projects a full white field brightness signal onto a standard sample bottle in a stationary state, and the image acquisition module monitors the original grayscale response value of the specular reflection area. The light intensity pre-regulation unit calculates the energy flow and adjusts the gain based on the current environment. The gain The calculation satisfies ,in, The system will adjust the gain based on the target grayscale value corresponding to 80% of the full-well charge capacity of the photosensitive chip. The brightness driving matrix, acting on the asymmetric spatial energy flow attenuation field, completes the amplitude correction, enabling the system to enter a stable detection state.

[0044] Example 6: Under the dynamic condition that the tested cylindrical container passes through the detection field of view at a linear velocity of 1.5 m / s, the labeling quality online detection system performs energy coordinate remapping based on motion trajectory prediction; the light intensity pre-regulation unit reads the encoder pulses of the conveyor line in real time to establish the instantaneous pose of the cylindrical surface in three-dimensional space, and then uses the pre-curved spatial homography matrix... Displacement vector generated by time step A composite operation is performed to calculate the dynamic conjugate coordinates of the specular reflection band on the surface of the tested cylindrical container on the spatial coding light projection module. In this composite operation, the system extracts the most convex generatrix of the cylinder that produces strong reflection when the tested cylindrical container moves along the conveyor belt in a pure straight line. The fixed-depth spatial tangent plane where it is located is set as a virtual two-dimensional calibration plane, so that the spatial translation of the three-dimensional rotationally symmetric surface is transformed into a two-dimensional linear vector transfer on the virtual tangent plane. This allows the two-dimensional homography matrix H, which represents the plane mapping relationship, to act directly on the dimension-reduced displacement vector s, realizing the equivalent tracking of the surface motion in the projection modulation domain. Specifically, the centroid of the saturated pixel group on the photosensitive chip of the image acquisition module is projected in real time to the physical control unit of the digital micromirror array using the projective transformation formula. This establishes the instantaneous center offset of the asymmetric spatial energy flow attenuation field in the modulation plane, thereby offsetting the geometric projection mismatch caused by the high-speed motion of the container.

[0045] To address the spatiotemporal energy accumulation deviation caused by the interaction between the projected light field and the object surface during high-speed flow, the light intensity pre-regulation unit superimposes a grayscale compensation operator based on the local reflectivity gradient distribution when constructing the brightness driving matrix. The system establishes a Gaussian-distributed attenuation weight function within a 12-pixel range around the resolved conjugate modulation site. The radial broadening coefficient of this weight function is determined by the reciprocal of the radius of curvature of the tested cylindrical container. This ensures that the optical energy flux projected onto the highly reflective zone is nonlinearly reduced based on the grayscale oversaturation ratio recorded in the initial reference image. This ensures that the local charge accumulation in the highly reflective area of ​​the modulated reflective image is suppressed to below 80% of the full-well charge capacity of the photosensitive chip within a single exposure window of 1μs to 999μs, maintaining the continuous characteristics of the spatial fringe topology at the highly reflective interface.

[0046] In the labeling quality online inspection system, to handle strong specular reflection background noise, the image acquisition module is equipped with a combination of linearly polarized filters with mutually orthogonal transmission directions on the light-incident side. The spatial coding light projection module has a first linearly polarized filter on its light-out side, and the fixed-focus optical lens of the image acquisition module has a second linearly polarized filter at its front end. The difference in polarization axis rotation angle between the first and second linearly polarized filters is constant at 90°. This allows secondary reflection stray light that does not conform to the projected polarization state in the tunable reflection image acquired by the image acquisition module to be filtered out through physical extinction. Under this optical path architecture, although the secondary stray light is effectively extinct, the physical stretching of the transparent label on the cylindrical curved surface induces a local stress birefringence effect. This causes the polarization plane of the initially linearly polarized strong specular reflection light to twist, resulting in approximately 10% of the initial strong reflection energy still being able to penetrate the second linearly polarized filter and approach the saturation threshold of the photosensitive chip. This physical-level approach addresses this residual high-gloss potential that is difficult to completely block due to the birefringence of the medium. The asymmetric spatial energy flow attenuation field generated by the polarization screening and light intensity pre-regulation unit ensures that the high-frequency distortion components extracted by the signal processing module in the spatial frequency domain are not interfered with by the saturation of the specular reflection light intensity, thus achieving physical enhancement of the microbubble defect characteristics in the labeling area. Under the dynamic transmission condition of the tested cylindrical container passing through the detection field of view at high speed, the hardware phase-locked loop reads the electrical pulse signal of the pipeline encoder and generates an exposure enable logic synchronized with the displacement vector of the cylindrical surface. The hardware phase-locked loop calculates the product of the encoder pulse frequency of the conveyor line and its corresponding spatial resolution, and sends a microsecond-level trigger command to the image acquisition module when the tested cylindrical container moves to the preset detection coordinates. This ensures that the image acquisition module accurately acquires one frame of modulated reflection image within every 1mm displacement step of the tested cylindrical container. This displacement feedback-based equidistant physical sampling mechanism eliminates the spatial phase broadening distortion of the image caused by the fluctuation of the conveyor line speed, providing a phase information source with a fixed spatial geometric reference for the signal processing module to reconstruct the three-dimensional morphological features of the labeling area.

[0047] Based on the displacement integral optical imaging model, the motion blur scale is equal to the product of the object's linear velocity and the exposure time. When the linear velocity of the measured cylindrical container is set to 1.5 m / s, in order to capture microbubble defects with a feature size of 15 μm, the system outputs a strobe enable signal with a pulse width of no more than 10 μs to the light source controller through a hardware phase-locked loop. This forces the global shutter exposure window of the image acquisition module to be compressed within this pulse width, thus limiting the physical displacement trailing image to the feature size of the microbubble defect. To address the random drift in container pose caused by high-speed transmission, the signal processing module extracts the low-frequency sinusoidal component basis phase distribution from two adjacent frames of modulated reflection images, calculates the local phase difference at the same pixel coordinates between the two frames, and determines the root mean square value of the local phase difference in the global pixel domain as the phase difference stability parameter. For cases where the phase difference stability parameter exceeds the set tolerance threshold, the signal processing module calculates the spatial coordinate offset based on the spatial phase-to-physical size mapping relationship. The calculation formula is: ,in, This represents the matrix of three-dimensional spatial coordinate offsets of the tested cylindrical container. This represents the extraction of the two-dimensional phase offset matrix of the low-frequency sinusoidal component. The equivalent spatial periodic matrix of the projected light field is calibrated on the surface of the cylindrical container under test, and the light intensity pre-tuning unit receives the spatial coordinate offset. Based on the pre-cured spatial homography matrix The coordinates of the digital micromirror array driving module of the spatially encoded optical projection module are remapped to perform subpixel-level coordinate compensation of the conjugate modulation site.

[0048] The embodiments of this application have been described above with reference to the accompanying drawings. Unless otherwise specified, the embodiments and features in the embodiments of this application can be combined with each other. This application is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms under the guidance of this application without departing from the spirit of this application and the scope of protection of this invention, and all of these forms are within the protection scope of this application.

Claims

1. An online labeling quality inspection system for detecting the labeling quality of a cylindrical container under test in a transmission state, characterized in that, include: A spatially encoded optical projection module is used to project a dual-frequency modulated light field with sinusoidal grayscale distribution characteristics onto the surface of a cylindrical container under test. The image acquisition module is used to acquire the modulated reflection image of the surface of the cylindrical container under test; The light intensity pre-regulation unit has its input end connected to the signal of the image acquisition module and its output end connected to the spatial coding light projection module. Based on the imaging optical center of the image acquisition module, the projection optical center of the spatial coding light projection module, and the transmission trajectory of the cylindrical container under test, the light intensity pre-regulation unit determines the conjugate modulation site of the specular reflection area on the spatial coding light projection module. When generating the brightness driving matrix of the dual-frequency modulation light field, the light intensity pre-regulation unit superimposes a static asymmetric spatial energy flow attenuation field in the local image domain where the conjugate modulation site is located. Before the photons interact with the surface of the cylindrical container under test, the incident brightness flux of the specular reflection area is reduced in advance, so that the charge accumulation of the highly reflective area in the modulated reflection image is maintained within the full-well charge capacity limit set by the photosensitive chip of the image acquisition module. The signal processing module is used to analyze the phase information of the modulated reflection image, extract the high-frequency distortion component that characterizes the labeling defects, and use the low-frequency wrapping phase distribution in the dual-frequency modulated light field as the order reference to complete the spatial phase unfolding of the high-frequency wrapping phase distribution, reconstruct the three-dimensional morphological features of the labeling area, and establish the labeling quality evaluation index by comparing the three-dimensional morphological features with the preset geometric model.

2. The labeling quality online inspection system according to claim 1, characterized in that, The system also includes a hardware phase-locked loop; this hardware phase-locked loop is connected to the displacement pulse output terminal of the transmission device and is used to lock the trigger pulse of the image acquisition module and the displacement pulse of the cylindrical container under test; the signal processing module calculates the spatial coordinate offset of the cylindrical container under test when the phase difference stability parameter deviates from the preset range based on the phase difference stability parameter between adjacent frame modulated reflection images, and performs coordinate compensation for the conjugate modulation site determined by the light intensity pre-regulation unit.

3. The online labeling quality detection system according to claim 1, characterized in that, The dual-frequency modulated optical field includes a high-frequency sinusoidal component and a low-frequency sinusoidal component, with a spatial frequency ratio of 10 to 30. The signal processing module uses the low-frequency phase field generated by the low-frequency sinusoidal component to establish the fringe order generated by the high-frequency sinusoidal component on the surface of the tested cylindrical container, and performs unwrapping operation on the phase field generated by the high-frequency sinusoidal component to eliminate the phase unwrapping deviation caused by the periodic jump of the fringe at the edge of the tested cylindrical container with high curvature.

4. The labeling quality online inspection system according to claim 1, characterized in that, The spatial coding light projection module includes a spatial light modulator composed of digital micromirror devices; the light intensity pre-regulation unit discretizes the two-dimensional coordinates of the conjugate modulation site on the spatial light modulator to generate a grayscale correction matrix corresponding to the static asymmetric spatial energy flow attenuation field, and superimposes the grayscale correction matrix onto the bottom driving sequence of the dual-frequency modulation light field, so that the projection energy of the mirror reflection area is non-uniformly distributed according to the curvature gradient of the surface of the tested cylindrical container.

5. The online labeling quality inspection system according to claim 1, characterized in that, When reconstructing the three-dimensional morphological features, the signal processing module extracts the energy feature distribution of the modulated reflection image in the spatial frequency domain. When there are microbubble defects on the label on the surface of the tested cylindrical container, the signal processing module identifies the local frequency offset signal in the modulated reflection image that deviates from the preset frequency threshold, thereby achieving physical isolation between the structural component corresponding to the microbubble defect and the background noise of the specular reflection area on the surface of the tested cylindrical container in the spatial frequency domain.

6. The online labeling quality detection system according to claim 1, characterized in that, The image acquisition module includes an industrial camera and a fixed-focus optical lens. The acquisition frequency of the industrial camera is not less than 100Hz. The main optical axis of the industrial camera and the projection axis of the spatial coding light projection module are arranged at an angle of 30 to 60 degrees in space, so that the imaging target surface of the image acquisition module and the modulation plane of the spatial coding light projection module form an optical conjugate mapping relationship, in order to establish a spatial sampling grid for the surface of the cylindrical container under test.

7. The online labeling quality inspection system according to claim 1, characterized in that, The signal processing module calculates the gradient deviation of the three-dimensional topographic features in the tangential direction of the surface. When the gradient deviation continues to exceed the preset slope threshold at the label edge, the signal processing module generates a judgment command characterizing the label edge lifting defect, and uses the local normal displacement of the reconstructed label area on the surface of the tested cylindrical container as a quantitative indicator of the defect degree to identify the label peeling state of the tested cylindrical container.

8. The online labeling quality inspection system according to claim 1, characterized in that, The image acquisition module is equipped with a linear polarization filter on the light-incident side; the signal processing module performs temporal smoothing filtering on 3 to 5 consecutive frames of modulated reflection images to reduce the signal-to-noise ratio loss caused by the mechanical displacement vibration of the tested cylindrical container during high-speed transmission and to suppress the random phase drift caused by ambient stray light, so as to improve the measurement repeatability of three-dimensional morphological features.

9. The labeling quality online inspection system according to claim 1, characterized in that, The system also includes a defective product separation unit; this defective product separation unit is connected to the output of the signal processing module and is used to receive the judgment instructions generated by the signal processing module. When the three-dimensional morphological features do not meet the preset quality standards, it generates a physical rejection action to remove the target cylindrical container from the main transmission path.

10. An online labeling quality detection method, used to implement the online labeling quality detection system of claim 1, characterized in that, Includes the following steps: Based on the imaging optical center of the image acquisition module, the projection optical center of the spatial coding optical projection module, and the transmission trajectory of the cylindrical container under test, the conjugate modulation site of the specular reflection area on the surface of the cylindrical container under test on the spatial coding optical projection module is determined. When generating the brightness driving matrix of the dual-frequency modulated light field, a static asymmetric spatial energy flux attenuation field is superimposed on the local image domain where the conjugate modulation site is located. Before the photon interacts with the surface of the cylindrical container under test, the incident brightness flux of the specular reflection area is reduced in advance, so that the charge accumulation in the highly reflective area of ​​the modulated reflection image is maintained within the full-well charge capacity limit of the photosensitive chip. The trigger pulse of the image acquisition module and the displacement pulse of the cylindrical container under test are locked by a hardware phase-locked loop to obtain the modulated reflection image of the surface of the cylindrical container under test. The phase information of the modulated reflection image is analyzed, the high-frequency distortion component characterizing the labeling defect is extracted, and the low-frequency wrapping phase distribution in the dual-frequency modulated light field is used as the order reference to complete the spatial phase unfolding of the high-frequency wrapping phase distribution and reconstruct the three-dimensional morphological features of the labeling area. Calculate the gradient deviation of the three-dimensional topographic features in the tangential direction of the surface. When the gradient deviation continues to exceed the preset slope threshold at the label edge, generate a judgment command characterizing the label edge lifting defect. Upon receiving the judgment instruction, if the three-dimensional morphological features do not meet the preset quality standards, the non-conforming product separation unit generates a physical rejection action to remove the target cylindrical container from the main transmission path.

Images